Networks play a central role in representing complex relationships among interconnected entities across diverse scientific domains. The increasing scale and complexity of real-world networks often make analytic tasks computationally expensive or intractable. Learning low-dimensional embeddings from high-dimensional and dependent network data has emerged as a powerful strategy, distilling essential structural information into manageable, interpretable, and computationally efficient representations. These embeddings are instrumental in facilitating downstream analyses and enhancing the practical utility of complex networks. In many data-driven scientific inquiries, researchers require not only accurate embedding estimates but also rigorous uncertainty quantification to ensure reliable inference and decision-making. Furthermore, data collected across varying conditions, time periods, or modalities often lead to the prevalence of multiple heterogeneous networks, yielding pressing needs for comparative and integrative analyses. This project will address these vital challenges by developing comprehensive methodologies for the estimation, inference, and integration of network embeddings. Additionally, it will generate broad educational impacts through research training opportunities for graduate and undergraduate students, innovations in curriculum development, and public engagement through outreach activities. This project will advance the statistical foundations of embedding le